Which AI trust, risk and security startups earn the most?

The ranking of top startups in the AI TRiSM market by revenue

Below is a table ranking the startups of this market by their current revenue scale. You can find our methodology at the end of this page.

If you want a deeper understanding of the market and its current dynamics, get our report on the AI TRiSM market.

Is AI TRiSM already producing real software businesses?

Clearly, AI TRiSM already has startups with meaningful recurring revenue and enterprise sales.

Arena and Alice sit well above most of the market on recurring sales, while Relyance AI, Holistic AI and Zenity form part of the next commercial tier. The exact order becomes less certain as company disclosures give way to estimates.

The catch is comparability. ARR, run-rate, annual revenue and third-party estimates all describe commercial scale differently. The market is clearly monetizing, but the public evidence gets much thinner once you move away from the biggest names.

Is Arena really a $100M AI trust and evaluation business already?

Yes, Arena reports >$100M of ARR, which puts it at the top of AI TRiSM startups by revenue.

Arena disclosed the figure while expanding from public model comparisons into real-world agent evaluation and alignment. Its commercial product now addresses questions enterprises actually pay to answer: whether models and agents behave as intended, follow instructions and complete work safely.

The metric still needs to be read correctly. Arena's figure is ARR, so it annualizes current recurring sales rather than reporting recognized full-year revenue.

Is Alice basically as big as Arena in AI TRiSM revenue?

Close: Alice reports ~$100M of ARR and sits very near Arena on the same recurring-revenue metric.

Both figures are ARR, so the comparison is unusually clean for this market. Alice also comes from a different starting point: it built a broad trust-and-safety business before pushing deeper into AI security.

In August 2026, Alice said it works with 8 of the 10 leading AI model labs and protects more than 3 billion people across major online platforms. That wider base helps explain how an adjacent trust business can become one of the largest AI TRiSM companies by recurring sales.

Does ARR make the biggest AI TRiSM startups look bigger than they are?

Only partly, because ARR is useful for AI TRiSM but should never be read as recognized full-year revenue.

Arena and Alice report ARR, while Galileo is ranked on a revenue run-rate. Those measures capture the current pace of recurring sales rather than revenue booked over a completed year.

The useful approach is simple: keep the metric attached to the number. That lets us compare commercial scale without pretending every private startup reports sales the same way.

The table below compares mid-sized AI TRiSM startups whose figures are annual revenue rather than ARR, mostly third-party estimates.

Company	Figure	Metric	Freshness
Aim Security	$10.6M	Annual Revenue	Aging · 26mo
Monitaur	$8.1M	Annual Revenue	Period not stated · collected this month
Aurascape	$6.4M	Annual Revenue	Fresh · 11mo
Knostic	$6.2M	Annual Revenue	Fresh · 14mo
ModelOp	$5.1M	Annual Revenue	Period not stated · collected this month
Lakera	<$5M	Annual Revenue	Fresh · 16mo
Openlayer	$4.8M	Annual Revenue	Aging · 29mo
ValidMind	$4.2M	Annual Revenue	Fresh · 14mo

Is Relyance AI already a big AI governance business?

Relyance AI looks like a meaningful AI governance business, although its revenue figure is still an estimate.

We rank Relyance AI on a ~$16.4M annual revenue estimate with no stated period. Its product scope is broad: AI inventory, data-flow discovery, governance, evaluation and runtime guardrails all sit in the same platform.

That breadth gives Relyance AI several enterprise buyers, from security and privacy teams to AI platform owners. The business looks substantial, but the exact revenue scale deserves less confidence than a recent company disclosure.

Are AI governance startups making serious money yet?

Yes, AI governance has already produced several startups with real commercial scale.

Holistic AI is ranked on a $13M annual revenue estimate, 2021.AI on $12.2M, Monitaur on $8.1M and ModelOp on $5.1M. Most of those figures are estimates, so they tell us more about order of magnitude than audited performance.

The products are also becoming more operational. Monitaur works with regulated enterprises, while ModelOp now pushes governance into policy enforcement across agent infrastructure. Customers are paying for software that manages AI after deployment instead of stopping at pre-launch documentation.

Are AI security startups making money faster than evaluation startups?

There is no clean winner yet between AI security and AI evaluation.

Arena and Galileo show that evaluation can monetize, while Alice and Zenity show the same for trust and security. The available figures use different metrics, so a clean speed comparison would be misleading.

Both paths are working. Security vendors can tap urgent risk budgets, while evaluation tools can become part of the AI development and production workflow. The public figures are too mixed to claim one model is winning across the market.

How much revenue does Zenity make from AI agent security?

Zenity has disclosed >$10M of annual revenue from its AI agent-security business.

When Zenity announced its latest financing in August 2026, the company said revenue had tripled in each of the past 2 years and was on track to triple again. Its customers are concentrated among large enterprises, and the product now covers agent discovery, governance and runtime protection.

That makes Zenity one of the clearest examples of agent security moving beyond pilots. Zenity did not say which period the figure covers, and >$10M is a floor rather than an exact number.

Has Galileo turned AI evaluation into a real business?

Yes, Galileo has built a real enterprise AI evaluation business with >$5M of revenue run-rate.

The Information reported that figure for January 2025, while Galileo said revenue had grown 834% since the beginning of 2024 and its enterprise customer count had quadrupled. The company has named customers and partners including Twilio, Comcast and HP.

The figure remains a run-rate rather than recognized full-year revenue. Even so, it shows that evaluation software can support meaningful enterprise contracts before a company reaches the very top of the market. Cisco completed its acquisition of Galileo in May 2026, so the company now sells inside a large group.

How big was Arize AI when Dynatrace bought it?

Arize AI was probably a bigger AI TRiSM business when Dynatrace bought it than its $12.4M annual revenue estimate for 2023 suggests.

That estimate describes the year ending December 2023 and is historical. Since then, Arize raised a $70M Series C in February 2025, expanded its evaluation and observability products, and became part of Dynatrace, which completed the acquisition in October 2026.

None of those events came with a new public revenue number, which shows how an old private-company estimate can lag the business it describes.

Is Fiddler AI still a standalone AI TRiSM business?

Not for long: Fiddler AI agreed in November 2026 to be acquired by CrowdStrike, so this AI TRiSM vendor is heading into a large security group.

The latest revenue figure for Fiddler AI is a $9.3M annual revenue estimate for the year ending December 2023. Since then, Fiddler has expanded into evaluations, agentic monitoring and runtime guardrails.

The acquisition makes that old estimate look even more dated, and no newer revenue figure is public. The $9.3M shows past sales, while the CrowdStrike deal shows what the business is worth to a buyer now.

Did Protect AI build a big business before Palo Alto Networks bought it?

Protect AI had real commercial traction, but the public revenue evidence points to a smaller business than the acquisition might suggest.

We rank Protect AI on a $5M annual revenue estimate for FY2025. Palo Alto Networks completed its acquisition of the company in July 2025 and said Protect AI's capabilities would become a cornerstone of Prisma AIRS.

Protect AI had built technology across model scanning, posture management, red teaming and runtime protection. The deal shows that a big buyer can pay for strong technology long before public revenue gets large.

Do famous AI security startups actually publish revenue?

Often, famous AI security startups publish customers, growth or usage while keeping current revenue private.

AIR Security disclosed >20 customers, while White Circle disclosed >1B API requests processed through May 2026. Noma Security has a $3.9M annual revenue estimate, but that estimate carries no date at all, not even the month it was collected.

Those facts show adoption, but they cannot replace a dated sales figure. That is why several prominent companies remain visible in the market without receiving a revenue rank.

Are European AI TRiSM startups making real money too?

Yes, European AI TRiSM startups are already generating meaningful revenue, even though much of the public evidence is estimated.

Holistic AI and 2021.AI are among the larger European governance vendors in our ranking. Lakera, the Zurich security startup that Check Point announced it would buy in September 2025, is on <$5M of reported annual revenue, while Saidot is on a ~$2.5M estimate.

The products span governance, compliance and security. Europe has clearly produced commercial AI TRiSM vendors, although no European figure in our ranking currently matches Arena or Alice on recurring sales.

Can we actually trust AI TRiSM startup revenue numbers?

Only when the metric, source and period stay attached to the number.

Arena and Alice disclose ARR directly, Zenity discloses annual revenue, and Galileo's run-rate comes from credible press reporting. Relyance AI, Holistic AI and many companies farther down the ranking rely on third-party estimates.

Freshness changes the picture too. Arize AI and Fiddler AI are both represented by historical estimates even though their businesses have moved on. Private-company revenue data is rarely perfect, so the useful question is how strong the evidence is behind each figure.

Are most AI TRiSM startups still small businesses?

Yes, most AI TRiSM startups are still small businesses, even though the market already has a few large outliers.

Openlayer is ranked on a $4.8M annual revenue estimate, ValidMind on $4.2M and Future AGI on $4.1M. Those are real businesses, but they are still far from the top of the ranking.

Enterprise security, governance and evaluation products often take time to pass procurement, integrate with existing systems and earn trust. The next test is whether more of these companies can turn early enterprise adoption into durable eight-figure revenue.

So, which AI trust, risk and security startups make the most money today?

Arena is the AI TRiSM startup that makes the most money right now, with Alice very close behind.

Relyance AI comes next, with a broader group of governance, evaluation and security vendors below them. The exact gaps become harder to judge once company disclosures give way to estimates and older figures.

The ranking also shows how young the market remains. Some leading numbers are recent company disclosures, others are run-rates or third-party estimates, and several famous companies have no rank because their public evidence is not current revenue.

The table below shows what we found.

Question	Answer
Is Arena the number one?	Yes. Arena ranks first on a company-disclosed ARR figure above $100M. Alice is close enough that the lead is narrow.
Is Alice already at major scale?	Yes. Alice reports ~$100M of ARR. Its broader trust-and-safety base helps explain that size.
Are governance startups making real money?	Yes. Holistic AI is estimated at $13M annual revenue. Governance software already supports meaningful commercial businesses.
Is agent security monetizing yet?	Yes. Zenity disclosed >$10M of annual revenue. Large enterprises are already paying for dedicated agent-security products.
Can evaluation software become a business?	Yes. Galileo reached >$5M of revenue run-rate. Enterprise evaluation has clearly moved beyond experimental budgets.
Can estimates outrank company disclosures?	Yes. Aim Security is ranked on a $10.6M estimate. Zenity sits just below it on a smaller figure that the company disclosed itself.
Can customer counts replace revenue?	No. AIR Security disclosed >20 customers but no revenue. Customers show traction, but they say nothing about sales.
Do acquisitions prove high revenue?	No. Protect AI was ranked on a $5M estimate. Strategic value can arrive before large public revenue.
Are old revenue figures still useful?	Yes, with caution. Arize AI's $12.4M estimate describes 2023. The company has clearly evolved since then.
Do famous startups sometimes stay unranked?	Yes. Some only disclose customers, usage or undated estimates. A revenue ranking needs a dated revenue figure.
Is AI TRiSM already mature?	Not yet. Many vendors remain in low single-digit millions. Arena and Alice are still major commercial outliers.